Download Case-Based Reasoning

Case-Based Reasoning
Introduction and Recent Developments
Ralph Bergmann, Klaus-Dieter Althoff, Mirjam Minor, Meike Reichle, Kerstin Bach
Case-based reasoning (CBR) is a sub-field of Artificial Intelligence that deals with experience-based problem solving. CBR has
its roots in different disciplines such as cognitive science, machine learning, and knowledge-based systems. Today, it is a well
established research field of its own, which produced a rich variety of specific methods, as well as applications implementing
those methods for particular tasks and domains. This paper gives a compact overview of CBR in general and further discusses
recent advancements in selected topics.
1
Introduction
Case-based Reasoning (CBR) is a well established research field
in Artificial Intelligence that involves the investigation of theoretical foundations [27], system development, and practical application building [10] of experience-based problem solving. The
core of every case-based problem solver is the case-base, which
is a collection of previously made and stored experience items,
called cases. A case-based problem solver solves new problems
primarily by reuse of solutions from the cases in the case base.
For this purpose, one or several relevant cases are selected. This
selection process is guided by one of the core assumptions behind
CBR, namely that similar problems have similar solutions. Once
similar cases are selected, the solution(s) from the case(s) are
adapted to become a solution of the current problem. Finally,
when a new (successful) solution to the new problem is found, a
new experience is made, which can be stored in the case-base to
increase its competence, thus implementing a learning behavior.
CBR has its roots in different disciplines, particularly cognitive science, machine learning, and knowledge based-systems, including knowledge representation and reasoning. CBR research
is also related to certain topics in information retrieval, data
bases, semantic web, and knowledge management. CBR is to a
large degree characterized by the fact that it combines methods
from different areas in AI in a particular manner for the purpose
of experience-based problem solving. Hence, there is a strong
focus on developing frameworks for certain types of problems
(e.g. diagnosis, planning, product recommendation, experience
management) as well as on implementing case-based systems for
certain application domains (e.g. for medicine). However, there
are also several CBR-related tasks for which other AI disciplines
do not already provide a solid methodological foundation. For
example, CBR research made significant original contributions
to the field of similarity modeling, similarity-based retrieval, and
adaptation. As several reviews of CBR exist [28, 32, 1], this
paper provides only a compact overview on CBR in general and
is then focussed on recent advancements in selected topics.
2
Knowledge and Reasoning in CBR
We now briefly describe CBR from the perspective of knowledge
representation and reasoning as this view shows the similarities
and differences to traditional methods used in AI.
2.1
Types of CBR
There are three main types of CBR that differ significantly from
one another concerning case representation and reasoning: structural, textual, and conversational CBR [10]. The idea underlying
the structural CBR approach is to represent cases according to
a common structured vocabulary, i.e. an ontology. Once this
vocabulary is defined, all cases are restricted to represent experience that can be expressed with this vocabulary. In the
various structural case representations, the describing features
of a case may be organized as flat attribute-value tables, in an
object-oriented manner, as graph structures, or by sets of atomic
formulas of a predicate logic language. The structural CBR approach is most widely used and will therefore be the focus of the
remainder of this survey.
In textual CBR, there is no common case structure, but cases
are represented as free text, i.e. strings [31]. This is very useful
in domains where large collections of know-how text documents
already exist and the intended user is able to make use of the
experience contained in the respective documents immediately.
In conversational CBR [5] cases aim at capturing the knowledge contained in customer/agent conversations. A case is represented through a list of questions that varies from one case to
another. There is no ontology and no standardized structure for
all the cases.
Today, structural CBR approaches make also use of features
from textual and conversational CBR. For example, textual cases
can be mapped to a structural representation by information extraction and by methods for automating ontological annotations.
Also, dialog components are used as part of CBR systems to implement specific strategies for user interaction.
2.2
CBR Cycle
Despite the many different appearances of CBR systems the
essentials of CBR are captured in a surprisingly simple and uniform process model (see Fig. 1), the CBR cycle proposed by
Aamodt and Plaza [3]. Several refinements have been proposed
for different purposes, e.g. for better addressing maintenance
issues in CBR. The CBR cycle consists of 4 sequential steps
organized around the knowledge of the CBR system. Problem
solving starts when a new problem (also called new case or query)
must be solved. First, the retrieve phase, selects one or several
similar cases from the case base. In the subsequent reuse phase
Page 1
the solutions contained in those cases are adapted according to
the query. In the revise phase, the solution determined so far is
verified in the real world and possibly corrected or improved, e.g.
by a domain expert. Finally, the retain phase takes the feedback
from the revise phase and updates the knowledge, particularly
the case base.
Problem
New
Case
RETRIEVE
Learned
Case
Retrieved
Case
Previous
Cases
RETAIN
New
Case
General
Knowledge
Tested/
Repaired
Case
REUSE
REVISE
Confirmed
Solution
Solved
Case
Suggested
Solution
Figure 1: The CBR Cycle according to Aamodt and Plaza.
Michael M. Richter proposed a unified view on the knowledge contained in a structural CBR application by introducing
different knowledge containers [46] thereby providing some additional structure to the general knowledge box in Figure 1. The
knowledge containers are the vocabulary, the case base, the similarity measure, and the adaptation knowledge.
2.3
Vocabulary and Case Representation
The vocabulary (which we call ontology today) is the basis of all
knowledge and experience representation in CBR. The vocabulary defines the information entities and structures (e.g. classes,
relations, attributes, data types) that can be used to represent
cases, similarity measures, and adaptation knowledge.
The case base is the primary form of knowledge in CBR. Traditionally, a case is considered as instance in the representation
space defined by the vocabulary, e.g. a vector in an attributevalue representation. However, more advanced representations
have been studied recently, such as hierarchical and generalized cases using case-specific constraints between attributes [54].
From a knowledge representation point of view, vocabulary and
case-base representations are a quite standard applications of AI
and database methods.
2.4
Similarity
The notion of similarity plays an important role in CBR since
cases are selected based on their similarity to the current problem. While early CBR approaches were usually restricted to
standard similarity measures (such as inverse Euclidean or Hamming distances), our current view is that the similarity measure
encodes important knowledge of the domain. Consequently, it
must be modeled as part of the knowledge acquisition process
during application development. Similarity is usually formalized
as a function sim : P ×P → [0, 1], which compares two problem
descriptions from P and produces a similarity assessment as a
real value from [0, 1]. A high value indicates a high similarity.
The semantics of such similarity measures can be defined via the
notions of preference and utility [9, 45]. For a new problem p
a case c1 = (p1 , s1 ) is preferred over a case c2 = (p2 , s2 ) (we
write c1 Âp c2 ) iff sim(p, p1 ) > sim(p, p2 ), since the similaritybased retrieval lists c1 before c2 . The exact similarity values are
usually not important, but just the preference order Âp that the
similarity function induces on the case base for the problem p.
The second observation is that this preference order should be
in line with the assessment of the utility of the solution of the
cases for solving the problem p during the reuse phase. Case
c1 should be preferred over c2 (c1 Âp c2 ) if the utility of s1 for
solving p is higher than the utility of s2 for solving p.
As a means for practical modeling of similarity functions, the
local-global-principle, first proposed by Richter (for details see:
[9, 45]) is widely used. Modeling similarity means decomposing
the similarity function according to the vocabulary. This decomposition is done in such a way that local similarity functions
for individual attributes model the preference according to the
attribute only. Local similarities are then aggregated into the
global similarity by an appropriate combination of the local similarity values. This aggregation also takes the different weights
of attributes into account. This approach has been further developed to support complex representations of vocabularies, including object-oriented representations [9].
2.5
Retrieve
In the retrieve phase of the CBR cycle, one or several cases
from the case base are selected, based on the modeled similarity.
In a nutshell, the retrieval task is defined as finding a small
number of cases from the case base with the highest similarity
to the query. Hence, this is a k-nearest-neighbor retrieval task
considering a specific similarity function. However, when the
case base grows, the efficiency of retrieval decreases, because an
increasing number of cases must be taken into account to find
the most similar case. Thus one sub-branch of CBR research
deals with methods that improve retrieval efficiency, e.g. by
using specific index structures such as kd-trees, case-retrieval
nets, or discrimination networks (an overview is given in [9]).
2.6
Reuse
When one or several similar cases have been retrieved, the solution (or other problem solving information) contained in these
cases is reused to solve the current problem. Reusing a retrieved solution can be quite simple if the solution is returned
unchanged as the proposed solution for the new problem. This is
mostly appropriate for classification tasks with a limited number
of solutions (classes) and a large number of cases. In such scenarios every potential solution is contained in the case base and
hence adaptation is usually not required. On the other hand,
for synthetic tasks (such as configuration or planning) solution
adaptation is necessary since the solution space clearly exceeds
the number of available cases or is even infinite.
Page 2
Several techniques for adaptation in CBR have been proposed so far (for a review on adaptation see [1]). The most basic
distinction between adaptation methods is whether transformational adaptation or generative adaptation is applied. Transformational adaptation relies on a set of adaptation rules or
operators that describe how differences in the problem lead to
required modifications in the solution. The present differences
between the new case and the retrieved case are analyzed, a
set of applicable transformations are selected, and the proposed
modifications to the solution are performed. On the other hand,
generative adaptation methods require a complete generative
problem solver that is able to solve problems based on general
knowledge, i.e., without using any cases at all. Not the solution but the problem solving traces from previous cases are then
reused to guide the generative problem solver to find a solution
to the new problem. This approach has its origin in derivational
analogy.
In any case, the acquisition of such explicit adaptation knowledge is a very difficult and time consuming task. Hence, most
practical CBR applications today try to avoid extensive adaptation for pragmatic reasons.
2.7
Revise
In this phase feedback related to the solution constructed so
far is obtained. This feedback can be given in the form of a
correctness rating of the result or in the form of a manually
corrected revised case. The revised case or any other form of
feedback enters the CBR system for its use in the subsequent
retain phase.
2.8
Retain
The retain phase is the learning phase of a CBR system. The
typical form of learning that occurs in a CBR system is learning by adding a revised case to the case base. Thereby, the
new problem solving experience becomes available for reuse in
future problem solving episodes. However, the continuous increase of the case base causes a utility problem as it continuously decreases retrieval efficiency. Explicit competence models
[47] have been developed that enable the selective retention of
cases.
In another branch of research, specific methods for learning
similarity measures have been developed. While early approaches
are restricted to learning feature weights [59] recent methods
address the more difficult problem of learning local and global
similarity functions [49, 17]. Very few approaches so far addressed the problem of learning adaptation knowledge [25, 19],
which is probably an issue of future interest, for example in the
context of the computer cooking contest [24].
3
Architectures, Frameworks, Tools
CBR can be found in both, architectures that are covering only
the CBR methodology or architectures in which CBR is one
methodology among others. The following paragraph presents
recent work on CBR architectures, frameworks, or tools focusing
solely on the CBR cycle (or on steps of it) within a knowledgeintensive application domain. The second paragraph describes
multi-agent-system architectures in which CBR and its underlying methodology play a significant role.
CREEK 1 is an architecture for knowledge-intensive CBR,
which has been developed at the Norwegian University of Science and Technology and focuses on open and weak-theory domains. Within CREEK, the case-based interpreter uses a threestep process to execute the retrieve, reuse and retain step. Each
step consists of activating relevant parts of the semantic network (vocabulary), generating and explaining derived information, and selecting a conclusion that conforms with the goal [2].
The (Java-based) implementation of the CREEK architecture is
the tool TrollCreek, which has been used to implement applications in the petroleum industry. jCOLIBRI 2 is an object-oriented
framework for building CBR systems that has been developed
by the Department of Software Engineering and Artificial Intelligence at Complutense University of Madrid. The underlying architecture of jCOLIBRI consists of two layers: The design
layer and the developer layer. The design layer provides tools
guiding users through the configuration process and explaining
the components’ behaviour within the application. The developer layer provides basic Java components, which are required
to create a CBR application from a developer’s point of view.
The architecture includes connectors to import different kinds
of data into a case base. Additionally, jCOLIBRI provides several extensions, among them components required for textual
CBR [44]. jCOLIBRI is used for teaching and research purposes
at several European universities and institutes. myCBR 3 [50]
is an open source CBR tool developed at the DFKI, which focuses on domain and similarity modeling for case retrieval. Its
intended use is in research and education, as well as for rapid
prototyping of applications. Since recently, models created with
myCBR can be used within applications built using the jCOLIBRI framework. IUCBRF 4 is also an open source framework
for CBR system development, which is implemented in Java and
developed at the Indiana University. The framework is designed
to facilitate fast and modular development of CBR systems as
well as providing a foundation for code sharing by those who are
developing CBR systems. On the commercial side the probably
most widely used tool in Europe is the Information Access Suite
(named e:IAS) from empolis5 . e:IAS uses CBR as its underlying methodology and provides a client-server based knowledge
provision component, the Knowledge Server, and a knowledge
modelling component, the Creator, for developing knowledge
management systems for a variety of commercial application
domains.
Another use of the CBR methodology is its inclusion in
multi-agent-systems. The combination of CBR and multi-agentsystems can either happen in the form of agent-enhanced CBR,
where intelligent software agents autonomously carry out tasks
of the CBR process [7] or in the form of CBR-enhanced agents,
where CBR is used to provide software agents with a reasoning capability [42]. Moreover, CBR can be combined with further technologies such as workflows [11]. The CAKE (Collaborative Agent-based Knowledge Engine) architecture, for instance, combines workflow technology, agent technology, and
1 http://creek.idi.ntnu.no/
2 http://gaia.fdi.ucm.es/projects/jcolibri/
3 www.mycbr-project.net
4 www.cs.indiana.edu/
sbogaert/CBR/
5 www.empolis.com
Page 3
structural CBR to select appropriate agents and workflows in
knowledge-intensive application domains using CBR [11]. Plaza
and Ontañón present the AMAL (Argumentation-based MultiAgent Learning) framework [40] for multi-agent learning, in which
CBR-agents increase prediction quality, efficiency, the range of
solvable problems and the range of accessible resources by arguing proposed solutions (learning to generate arguments and
counterarguments). Within AMAL, software agents are able to
advance their reasoning capabilities by learning knowledge and
preference relations from experience. The distribution of tasks
and knowledge by means of a Collaborative Multi-Expert-System
has been introduced by Bach et al. [8] presenting the SEASALT
(Sharing Experience using an Agent-based System Architecture
LayouT) architecture. This architecture comprises, among other
components, the so-called Knowledge Line, a number of intelligent CBR-based agents, which enable distributed knowledge
management. Further on, these agents’ case bases are again
maintained by agents (Case Factory), thus combining the approaches of CBR-enhanced agents and agent-enhanced CBR.
4
CBR and the Future Internet
The development of the future internet is affected by two major
factors: semantics and collaboration. Since the amount of information available on the internet is rapidly increasing, a more
meaningful, semantic description of the available content is necessary in order to tackle the problem of information overflow.
This need is addressed by the Semantic Web. At the same time
the internet’s content no longer mainly consists of editorial content and static private homepages, but has also become a social
medium. Today the internet is not only used for information
gathering, but also for social networking and sharing personal
views, experiences and media. This development is usually subsumed by the term Web 2.0. Both these new fields, the Semantic Web as well as the Web 2.0, benefit from the application of
case-based reasoning.
Two of the most influencing developments of the Semantic Web are the resource description language RDF (Resource
Description Framework) and the knowledge representation language OWL (Web Ontology Language), which is based on RDF.
Already before the development of RDF and OWL, XML has
been used as a case representation within the case-based reasoning community. There is a notable similarity between the
ontologies developed within semantic applications and the representation of cases in structural case-based reasoning. Due to
this similarity RDF and OWL both lend themselves to be used as
case representation languages and thus expand the possibilities
of case-based reasoning within the general WWW. Bergmann
and Schaaf [12] illustrate the technological and methodological
similarities between ontologies and structured case-based reasoning and describe the synergies that can be reached by merging
both approaches. A similar approach is presented by Chen and
Wu [16], who describe an RDF based Case Markup Language
called CaseML. CaseML offers a domain-independent case ontology and also aims to make case-based reasoning available within
the Semantic Web. Aktas et al. [6] also use RDF for case representation in their system SERVOGrid. Here it is embedded
in a conversational case-based reasoning system that aids scientists in finding resources such as program code or data that are
needed to solve a specific task by assisting them in describing
the necessary resources using meta data. Another case-based
reasoning architecture that is making heavy use of Semantic
Web technologies is the jCOLIBRI framework. Here OWL is being used as the case interchange language and it is planned to
advance the already distributed framework towards an architecture consisting of Semantic Web Services (SWS) where problem
solving methods are represented as Web Services [43]. In order to use these services the whole case-based reasoning process
is decomposed into single tasks, which are then carried out by
according Web Services. D’Aquin et al. [20] also aim to integrate case-based reasoning with Semantic Web technologies,
but this time using OWL for representing adaptation knowledge
and applying OWL reasoning in order to carry out case-based
reasoning’s retrieval and adaptation steps.
Within Web 2.0 the area of research being most related to
case-based reasoning is the collaborative filtering approach. Already in 2001 Hayes et al. [26] described the close relation between collaborative filtering and CBR and how these can benefit
from each other. Since then combinations of CBR and collaborative filtering have been presented in several papers. O’Sullivan
et al. [53] use collaborative filtering profiles as cases in a TV recommender system. There are also other combinations possible
such as in the work by Chedrawy et al. [15], where CBR adaptation techniques are used to refine the results of collaborative
filtering or in the work by Aguzzoli et al. [4], in which collaborative filtering is used to assess the similarity between songs
in a CBR system creating custom music compilations (CoCoA).
Guo et al. [23] further extend the application of CBR on collaborative filtering data – this time within a recommender system
in e-commerce – by additionally adding a social trust model to
improve prediction accuracy. A slightly different approach has
been presented in the recent paper by Briggs and Smyth [14],
which illustrates the idea of a community based web search that
uses the results of previous web searches of similar users in order
to improve web search results.
Most of the approaches presented above rely on already
structured or otherwise prepared cases. Given the assumption
that the cases used in CBR systems are usually representation
of experiential knowledge, the next logic step would be to try to
directly access and (re)use the experiential knowledge enclosed
in Web 2.0 platforms. Plaza [41] introduces the EDIR cycle for
this purpose, which describes four processes (Express - Discover
- Interpret - Reuse) for reusing experiential knowledge directly
from the web.
5
Procedural Knowledge in CBR
Procedural knowledge plays an important role in many companies. Experience in the form of procedures has found its way
into CBR mainly in two fields: case-based planning (CBP) and
workflow-oriented CBR.
In CBP, a plan describes procedural knowledge per se. The
single steps of a plan as well as the sequential or hierarchical
order of the steps contain a lot of experience that can be addressed by CBR. Instead of planning from scratch, CBP focusses
mainly on retrieval and adaptation of plans. CBP has a history
of nearly 20 years, which has been summarized in the literature
[48, 1]. As it is still a vivid research area, we will concentrate
Page 4
here only on some recent trends that have emerged in the past
three years. The utility of cases for solving planning problems
has been investigated in order to improve the performance of
planning [21]. Some work has been done on planning under
real-time constraints [39, 52]. By considering personalized plan
execution [29], an existing knowledge planning approach (plans
are search procedures) has been extended in order to rank the
results of applying the plan (search results) according to user
preferences.
Workflows are “the automation of a business process, in
whole or part, during which documents, information or tasks are
passed from one participant to another for action, according to a
set of procedural rules” [18]. In a medical domain, for instance,
such tasks might be the diagnosis and treatment steps for a
patient, while the procedural rules determine the control flow
for the execution of the tasks. Workflow-oriented CBR deals
with knowledge on how to model, apply, and adapt procedures
in the form of workflows. The reuse of workflow templates is
widely spread in recent commercial systems. Before a workflow is enacted, a new workflow instance is derived from the
workflow template. CBR is a means to go beyond this kind of
assistance. First case-based approaches exist that give sophisticated modeling support for workflows. Madhusudan et al. [33]
support the incremental modeling of workflows by a similaritybased reuse of the workflow templates. A process repository
with workflows and workflow snippets is used by an HTN planner that employs an inexact, graph-based matching. Leake and
Morwick [30] evaluate the execution paths of past workflows in
order to support user extension of workflows that are under construction. The application of workflows is supported by CBR
as well. Case-based retrieval functionality is employed to select
suitable workflows for a certain situation [11] or for knowledgeintensive tasks of a knowledge worker [56]. The adaptation of
workflows at execution time is also a recent research topic in
CBR. It can not yet be performed automatically but it can benefit from CBR approaches. Conversational CBR has been applied
in the tool CBRFlow [58] to guide the user in adapting a workflow to changing circumstances. The changes between revisions
of past workflows have been reused in the tool CAKE [34] in
order to generate suggestions for workflow adaptations to the
user.
6
CBR Applications
During the past twenty years, many CBR applications have been
developed, ranging from prototypical applications build in research labs to large-scale fielded applications developed by commercial companies. Application areas of CBR include help-desk
and customer service, recommender systems in electronic commerce, knowledge and experience management, medical applications and applications in image processing, applications in law,
technical diagnosis, design, planning, as well as applications in
the computer games and music domain. CBR applications are
documented in part in the conference proceedings of the regular ICCBR/ECCBR conferences as well as on the Industry Day
included in these conferences.
Watson [57] reports that more than 130 major companies
had been fielding CBR applications worldwide till 1997. Early
prominent fielded applications in engineering and technical do-
mains included CLAVIER (configuration of autoclave loadings),
Cassiopee (troubleshooting of CFM56-3 engines in the Boeing
737) and ICARUS (locomotive diagnostics). Help-desks and customer support are two of the most obvious applications of CBR
and many such applications had already been implemented by
1999 [51].
An important application area that also specifically addresses
the method-oriented view of CBR is experience management or,
if considered from a more general perspective, knowledge management. A well-known application example is SQUAD, which
deals with corporate knowledge management for supporting software quality control. While Tautz [55] describes how CBR is
used as part of a more general method for developing experience management applications, Nick [37] provides details on
a number of real-life applications of CBR for experience management tasks. Interesting real-life applications on experience
management include the variation reduction adviser [36], and
the connection machine [22], an approach for collective expertise development.
Bichindaritz [13] gives an overview of the use of CBR in
the health sciences. Bergmann at al. [10] describe more than
two dozens of real-life applications of CBR developed using the
INRECA methodology and tools. Recent applications on autonomous systems include applications of CBR for ambient intelligence [38] and self-healing [35]. Mantaras et al. [1] give an
overview on emergent applications of CBR in domains like music (e.g., generating expressive musical performances), poetry
generation, computer games (script adaptation, character control, planning in strategy games), molecular biology (speed-up
of crystal growth process for proteins) and support for spatial
reasoning in geographical information systems.
7
Conclusion
By now it is difficult to exhaustively report on a few pages on
the current state of the art in CBR, because the number of CBR
approaches and applications developed up to now has become
quite large. Apart from the broad applicability of CBR, one reason for this is that there is meanwhile a significant number of
CBR research groups and commercial companies, which develop
CBR methods, software components, and applications on a regular basis. Another reason is that already in the 1990s various
CBR researchers pointed out that CBR is not only a technology
but also a (process oriented) method. Since CBR researchers
are experts in processing experience, detailing, adapting, and
improving the CBR method for various innovative and challenging application fields and tasks has been an important research
area in the CBR community from its very beginning. This also
explains why the combination of CBR with various other technologies within a great bandwidth of applications has become
increasingly attractive for researchers as well as business professionals. Especially for these integrated approaches it is not
always made explicit that there is indeed CBR inside.
References
[1] Case-based reasoning commentaries. The Knowledge Engineering Review, 20(3), 2006.
Page 5
[2] A. Aamodt. Knowledge-intensive case-based reasoning in creek.
In Advances in Case-Based Reasoning, ECCBR 2004, LNAI 3155,
pages 1–15. Springer, 2004.
[19] S. Craw, J. Jarmulak, and R. Rowe. Learning and applying casebased adaptation knowledge. In Case-Based Reasoning Research
and Development, ICCBR 2001. Springer, 2001.
[3] A. Aamodt and E. Plaza. Case-based reasoning: foundational
issues, methodological variations, and system approaches. AI
Communications, 7(1):39–59, 1994.
[20] M. d’Aquin, J. Lieber, and A. Napoli. Case-based reasoning
within semantic web technologies. In Proceedings of the 12th
International Conference on Artificial Intelligence: Methodology, Systems, and Applications, AIMSA2006, pages 190–200.
Springer, 2006.
[4] S. Aguzzoli, P. Avesani, and P. Massa. Collaborative case-based
recommender system. In S. Craw and A. Preece, editors, Advances in Case Based Reasoning, ECCBR 2002, pages 460–474.
Springer, 2002.
[5] D. Aha, L. A. Breslow, and H. Munoz-Avila. Conversational
case-based reasoning. Applied Intelligence, 2001.
[6] M. S. Aktas, M. Pierce, G. C. Fox, and D. Leake. A web based
conversational case-based recommender system for ontology
aided metadata discovery. In Proceedings of the 5th IEEE/ACM
International Workshop on Grid Computing, GRID2004, pages
69–75. IEEE Computer Society, 2004.
[7] K.-D. Althoff, K. Bach, J.-O. Deutsch, A. Hanft, J. Mänz,
T. Müller, R. Newo, M. Reichle, M. Schaaf, and K.-H. Weis. Collaborative multi-expert-systems – realizing knowlegde-productlines with case factories and distributed learning systems. In 3rd
Workshop on Knowledge Engineering and Software Engineering
(KESE 2007), 2007.
[8] K. Bach, M. Reichle, and K.-D. Althoff. A domain independent
system architecture for sharing experience. In A. Hinneburg,
editor, Proceedings of LWA 2007, Workshop Wissens- und Erfahrungsmanagement, pages 296–303, 2007.
[9] R. Bergmann. Experience Management - Foundations, Development Methodology, and Internet-Based Applications. LNAI
2432. Springer, 2002.
[10] R. Bergmann, K.-D. Althoff, S. Breen, M. Göker, M. Manago,
R. Traphöner, and S. Wess. Developing industrial case-based
reasoning applications: The INRECA methodology. LNAI 1612.
Springer, 2nd Edition, 2003.
[11] R. Bergmann, A. Fressmann, K. Maximini, R. Maximini, and
T. Sauer. Case-based support for collaborative business. In
Advances in Case-Based Reasoning, ECCBR 2006, LNAI 4106,
pages 106–120. Springer, 2006.
[12] R. Bergmann and M. Schaaf. Structural case-based reasoning
and ontology-based knowledge management: A perfect match?
Journal of Universal Computer Science, 9(7):608–626, 2003.
[21] T. de la Rosa, A. Garcı́a Olaya, and D. Borrajo. Using cases
utility for heuristic planning improvement. In Case-Based Reasoning Research and Development, ICCBR 2007, LNAI 4726,
pages 137–148. Springer, 2007.
[22] M. H. Göker, C. Thompson, S. Arajärvi, and K. Hua. Connecting people with questions to people with answers. Künstliche
Intelligenz 4/07, pages 23–27, 2007.
[23] Y. Guo, G. Deng, G. Zhang, and C. Luo. Using case-based
reasoning and social trust to improve the performance of recommender system in e-commerce. In Proceedings of the Second
International Conference on Innovative Computing, Informatio
and Control, ICICIC 2007, page 484. IEEE Computer Society,
2007.
[24] A. Hanft, N. Ihle, K. Bach, and R. Newo. Cookiis – competing
in the first computer cooking contest. This Volume.
[25] K. Hanney and M. Keane. Learning adaptation rules from a case
base. In Proceedings of the 3th European Workshop on CaseBased Reasoning, LNAI 1168, pages 179–192. Springer, 1996.
[26] C. Hayes, P. Cunningham, and B. Smyth. A case-based reasoning
view of automated collaborative filtering. In D. Aha and I. Watson, editors, Case-Based Reasoning Research and Development,
ICCBR 2001, pages 243–248. Springer, 2001.
[27] E. Hüllermeyer. Case-Based Approximate Reasoning. Springer,
2007. See book review in this volume.
[28] J. Kolodner. Case based reasoning, 1993.
[29] D. Leake and J. Powell. Knowledge planning and learned personalization for web-based case adaptation. In Advances in CaseBased Reasoning, ECCBR 2008, LNAI 5239, pages 284–298.
Springer, 2008.
[30] D. B. Leake and J. Kendall-Morwick. Towards case-based support for e-science workflow generation by mining provenance. In
Advances in Case-Based Reasoning, ECCBR 2008, LNAI 5239,
pages 269–283. Springer, 2008.
[13] I. Bichindaritz. Role and significance of case-based reasoning in
the health sciences. This Volume.
[31] M. Lenz and K. Ashley, editors. Proceedings of the AAAI98
Workshop on Textual Case-Based Reasoning. AAAI Press, 1998.
[14] P. Briggs and B. Smyth. Provenance, trust, and sharing in peerto-peer case-based web search. In Advances in Case-Based Reasoning, ECCBR 2008, LNAI 5239, pages 89–103. Springer, 2008.
[32] M. Lenz, Bartsch-Spörl B., H.-D. Burkhard, and S. Wess, editors. Case-Based Reasoning Technology: From Foundations to
Applications. Springer, 1998.
[15] Z. Chedrawy, S. Sibte, and R. Abidi. Intelligent knowledge sharing strategy featuring item-based collaborative filtering and case
based reasoning. In Proceedings of the 5th International Conference on Intelligent Systems Design and Applications, ISDA2005,
pages 67–72. IEEE Computer Society, 2005.
[33] T. Madhusudan, J. L. Zhao, and B. Marshall. A case-based
reasoning framework for workflow model management. Data
and Knowledge Engineering, 50:87–115, 2004.
[16] H. Chen and Z. Wu. Caseml: A rdf-based case markup language for case-based reasoning in semantic web. Technical Report 4/2003, Norwegian University of Science and Technology,
Department of Computer and Information Science, 2003.
[17] W. Cheng and E. Hüllermeier. Learning similarity functions from
qualitative feedback. In Advances in Case-Based Reasoning, ECCBR 2008, LNAI 5239. Springer, 2008.
Workflow management
[18] Workflow Management Coalition.
coalition glossary & terminology. Retrieved on Dec 16, 2008
from
http://www.wfmc.org/View-document-details/WFMCTC-1011-Ver-3-Terminology-and-Glossary-English.html, 1999.
[34] M. Minor, A. Tartakovski, and R. Bergmann. Representation
and structure-based similarity assessment for agile workflows. In
Case-Based Reasoning Research and Development, ICCBR 2007,
LNCS 4626. Springer, 2007.
[35] S. Montani and C. Anglano. Case-based reasoning for autonomous service failure diagnosis and remediation in software
systems. In Advances in Case-Based Reasoning, ECCBR 2006,
LNAI 4106, pages 489–503, 2006.
[36] A. P. Morgan, J. A. Cafeo, K. Godden, R. M. Lesperance, A. M.
Simon, D. L. McGuinness, and J. L. Benedict. The general motors variation – reduction adviser: A deployed system for experience management. Künstliche Intelligenz 4/07, pages 27–29,
2007.
Page 6
[37] M. Nick. Experience Maintenance Through Closed-Loop Feedback. PhD thesis, University of Kaiserslautern, 2005.
[38] M. Nick, S. Schneickert, J. Grotepaß, and I. Heine. Checkmate.
Künstliche Intelligenz 4/07, pages 34–37, 2007.
[39] S. Ontañón, K. Mishra, N. Sugandh, and A. Ram. Case-based
planning and execution for real-time strategy games. In CaseBased Reasoning Research and Development, ICCBR 2007, LNAI
4726, pages 164–178. Springer, 2007.
[40] S. Ontañón and E. Plaza. Learning and joint deliberation through
argumentation in multi-agent systems. In Proc. AAMAS-2007,
volume 13 N. 3, pages 971–978, 2007.
[41] E. Plaza. On reusing other people’s experiences. This Volume.
[42] E. Plaza and S. Ontañón. Learning collaboration strategies for
committees of learning agents. Journal of Autonomous Agents
and Multi-Agent Systems, 13 N. 3:429–461, 2006.
[43] J. A. Recio-Garcı́a, B. Dı́az-Agudo, and P. A. González-Calero.
A distributed cbr framework through semantic web services. In
Research and Development in Intelligent Systems XXII, pages
88–101. Springer, 2005.
[44] J. A. Recio-Garcı́a, B. Dı́az-Agudo, A. Sánchez-Ruiz, and P. A.
González-Calero. Lessons learnt in the development of a cbr
framework. In Proccedings of the 11th UK Workshop on Case
Based Reasoning, pages 60–71. CMS Press,University of Greenwich, 2006.
[45] M. Richter. Foundations of similarity and utility. In Proceedings
of the 20th International Florida Artificial Intelligence Research
Society Conference (FLAIRS 2007). AAAI Press, 2007.
[58] B. Weber, W. Wild, and R. Breu. CBRFlow: Enabling Adaptive Workflow Management Through Conversational Case-Based
Reasoning. In Advances in Case-Based Reasoning, ECCBR 2004,
LNCS 3155, pages 434 – 448. Springer, 2004.
[59] D. Wettschereck and D.W. Aha. Weighting features. In CaseBased Reasoning Research and Development, Proc. ICCBR-95,
LNAI 1010, pages 347–358. Springer, 1995.
Contact
Prof. Dr. Ralph Bergmann
University of Trier
Department of Business Information Systems II
54286 Trier, Germany
Tel.: +49 (0)651 201 3876
Email: [email protected]
Prof. Dr. Klaus-Dieter Althoff
University of Hildesheim
Intelligent Information Systems Lab
Marienburger Platz 22, 31141 Hildesheim
Tel.: + 49 (0)5121-883 751
Email: [email protected]
Bild
[46] M. M. Richter. The knowledge contained in similarity measures.
Invited Talk on the ICCBR-95, 1995.
[47] B. Smyth and E. McKenna. Competence models and the maintenance problem. Computational Intelligence, 17(2):235–249,
2001.
[48] L. Spalazzi. A survey on Case-Based Planning. Artificial intelligence review, 16:3–36, 2001.
[49] A. Stahl. Learning similarity measures: A formal view based on
a generalized cbr model. In Case-Based Reasoning Research and
Development, ICCBR 2005, LNAI 3620. Springer, 2005.
Bild
[50] A. Stahl and T. Roth-Berghofer. Rapid prototyping of cbr applications with the open source tool mycbr. This Volume.
[51] M. Stolpmann and S. Wess. Optimierung der Kundenbeziehungen mit CBR-Systemen – Intelligente Systeme für E-Commerce
und Support. Addison Wesley, 1999.
[52] N. Sugandh, S. Ontañón, and A. Ram. Real-time plan adaptation
for case-based planning in real-time strategy games. In Advances
in Case-Based Reasoning, ECCBR 2008, LNAI 5239, pages 533–
547. Springer, 2008.
[53] D. O’ Sullivan, D. Wilson, and B. Smyth. Improving case-based
recommendation: A collaborative filtering approach. In Proceedings of the 6th European Conference on Advances in Case-Based
Reasoning, ECCBR 2002, pages 278–291. Springer, 2002.
[54] A. Tartakovski. Reasoning with generalized cases. This Volume.
[55] C. Tautz. Customizing Software Engineering Experience Management Systems to Organizational Needs. PhD thesis, University of Kaiserslautern, 2001.
[56] L. van Elst, F. R. Aschoff, A. Bernardi, H. Maus, and S. Schwarz.
Weakly-structured Workflows for Knowledge-intensive Tasks:
An Experimental Evaluation. In 12th IEEE International Workshops on Enabling Technologies, WETICE 2003, pages 340–345.
IEEE Computer Society, 2003.
Bild
Ralph Bergmann is full professor at the University of Trier and is directing a research
group on business information systems. He
studied computer science at the University of
Kaiserslautern and finished his doctoral dissertation on hierarchical case-based planning
in 1996. In 2001 he received the postdoctoral
degree (Habilitation) with a thesis on experience management. His main interests include case-based reasoning, experience management, machine learning, semantic technologies including
semantic
Klaus-Dieter
Althoff
is fullworkflow
professorreasonat the
ing.
University
of Hildesheim and is directing the
Intelligent Information Systems Lab. He
wrote his Phd thesis on an architecture for
knowledge-based technical diagnosis at the
University of Kaiserslautern, where he also
received the postdoctoral degree (Habilitation) with a thesis on the evaluation of casebased reasoning systems. His main interests
include knowledge-based systems, with a focus on case-based reasoning, agent technology, experience
Mirjam
Minor ismanagement,
a postdoc atand
themachine
Univerlearning.
sity of Trier, Germany, in the research group
on business information systems. She studied computer science at the Humboldt Berlin,
and received her PhD degree there in 2006.
Her main research interests are in agile workflow technology, case-based reasoning, experience & knowledge management, and multiagent systems.
[57] I. Watson. Applying case-based reasoning: techniques for enterprise systems. Morgan Kaufmann Publishers Inc., 1998.
Page 7
Bild
Bild
Meike Reichle studied International Information Management with Computer Science
and History as minors at the University of
Hildesheim. She graduated in 2007 and is
since working in a PhD position at the University of Hildesheim’s Intelligent Information Systems Lab. Her main research interests include the combination of case-based
reasoning and machine learning techniques,
multi-agent systems, knowledge and experience maintenance, and natural language processing. Bach studied Information ManageKerstin
ment and Information Technology at the University of Hildesheim from 2001 to 2007.
She graduated in 2007 and is since working in a PhD position at the University of
Hildesheim’s Intelligent Information Systems
Lab. Her main research interests include
intelligent information systems, agent-based
systems, travel medicine, case-based reasoning, textual case-based reasoning, knowledge
discovery and similarity.
Page 8